Trajectory Modeling with Latent Groups: Potentials and Pitfalls

Abstract

The goal of this review is to provide a non-technical overview of trajectory modeling with latent groups. The paper will provide guidance to researchers seeking to use these methods with concrete recommendations on steps to follow and potential challenges. Trajectory modeling with latent groups is a quickly evolving field with new findings on best practices. Without clear reporting of how models were created, they risk providing conflicting conclusions and irreproducible results. Trajectory models with latent groups provide a powerful way of understanding patterns of change in health status, behaviors, and other outcomes. Their growing use in epidemiology has tremendous potential for understanding life course exposures, disease progression, and patterns of health behavior over time. However, without careful consideration of how models are structured, they are at risk of identifying artificial patterns and subgroups.